SimTrip: Modelling trip similarity for travel recommendations

Author Bios (50 Words for each Author)

Ankur Mawandia works as a part of data analytics team, Business Intelligence Unit at Amadeus Labs Bangalore. He has completed his M.S in Artificial Intelligence from Technical University of Kaiserslautern, Germany. His research interests include choice modeling, prediction systems, and activity recognition.

Deepak Sunil works as a Business Analyst for Business Intelligence Unit in aviation sector at Amadeus Labs Bangalore

Shashwat Sharma works as a Data Engineer for Business Intelligence Unit in aviation sector at Amadeus Labs Bangalore

Deepak Sunil works as a Data Engineer for Business Intelligence Unit in aviation sector at Amadeus Labs Bangalore

Abstract (150 Words)

SimTrip: Modelling trip similarity for travel recommendations

Proposing presonalized travel recommendations using trip similarity and reinforcement learning

Abstract

In this paper, we present a self-adaptive model to make personalized trip recommendations. We train our model on 100 city pair locations using a heuristic approach for city pair similarities. We find trip similarity on any origin-destination combinations, allowing us to make personalised relevant recommendations. We use publically available economic, geographic, climate and demographics data as an input to our model. The similarity score is updated on user feedback to capture trend and seasonality for model updates. We discuss the calibration methods to tune the recommendation model and suggest evaluation techniques. We also present use case scenarios for our model.

This document is currently not available here.

Share

COinS
 

SimTrip: Modelling trip similarity for travel recommendations

SimTrip: Modelling trip similarity for travel recommendations

Proposing presonalized travel recommendations using trip similarity and reinforcement learning

Abstract

In this paper, we present a self-adaptive model to make personalized trip recommendations. We train our model on 100 city pair locations using a heuristic approach for city pair similarities. We find trip similarity on any origin-destination combinations, allowing us to make personalised relevant recommendations. We use publically available economic, geographic, climate and demographics data as an input to our model. The similarity score is updated on user feedback to capture trend and seasonality for model updates. We discuss the calibration methods to tune the recommendation model and suggest evaluation techniques. We also present use case scenarios for our model.